Advertising

HTML5 Banners

We had to go through special approval to get HTML5 banners supported on our account.

Steam Store Page

Once we did, we ran a campaign for Steam with audiences:

• Media & Entertainment -> Hardcore Gamers (Affinity Audience)
• space games (Custom intent audience: auto-created)

101 clicks, 30,755 impressions, $50.75 ... no conversions

Direct Download Page

I then switched to Direct with the same audiences 52 clicks, 54,446 impressions, $35.11 ... no conversions

We turned on the AdWords API integration on 11/12/2019 using Direct link with audiences:

• Media & Entertainment -> Hardcore Gamers (Affinity Audience)
• space games (Custom intent audience: auto-created)

44 clicks, 35,501 impressions, $26.06 ... no conversions

11/14/2019 Today I have switched up the audiences and running some experiments on both Steam and Direct ads. Steam was 40-18, 403 in wishlist.

96 clicks, 52,995 impressions, $39.35 ... no conversions ... 1 added to wishlist on Steam

Reset

New RiseTVP.com home page 11/19 ... we ran some advertising ... Steam we got a good review from User:corgi so we would like to see if that makes a difference too!

Facebook

Not sure where the documentation went but by far the best bang for our buck has been with Facebook. We have been integrated since the beginning of 2020 with Facebook Ads SDK. Watching LogRocket, the Google Adwords respondents appeared to be mostly bots... maybe even suspiciously operated by Google with the express purpose of costing people more money! Facebook respondents on the other hand interacted with the page. Facebook also assigned a sales engineer to us for a couple months that helped us get Lookalike audiences up and running which cut our CPC rate to basically cut in half.

6/29/2020 Update

Looking back at the analytics from 4/23/2020 to today I noticed a dip on Thursdays (yesterday we only got 2 new players for $16!). So I ran some numbers and found that the weekends work best for us and Tuesday/Thursday are the worst but not by as much as I thought. I even threw some standard deviation computations in there and ultimately we are shifting about 30% of our budget from Thursday to other days.

The Great ChatGPT Advertising Experiment of 2026

Setup and review: 23 October 2026. Can ChatGPT advertising find people who enjoy becoming capable operators, supplying settlements, and contributing to a persistent world? That is the experiment. Its purpose is to acquire players who will feel at home in Rise and return, with clicks serving as an intermediate measure.

Current setup — 23 October 2026

The current experiment uses one English ad and one shared US$25 average daily campaign budget. This supersedes the original United States-only baseline below and the brief expansion to 26 countries. Account and campaign settings were checked again on 23 October; this is a configuration record, not a claim of successful delivery or player acquisition.

Item	Current saved configuration
Campaign	Rise: The Vieneo Province; active; Clicks objective. The campaign was originally named Risetvp campaign.
Active creative	One English ad, Vieneo Colony, with approved copy: “Rise: Fly. Supply. Build.” / “Master atmospheric flight. Supply colonies. Shape a persistent world.” The nine manually translated ads are archived.
Selected countries	United States, Canada, United Kingdom, Ireland, Australia, New Zealand, Singapore, and South Africa. Country selection does not establish a viewer's language or prove delivery eligibility in every selected country.
Budget and schedule	US$25 average daily, total for the campaign. America/Chicago timezone; no end date or total campaign cap. No increase was made for additional countries or translations.
Public Brand Name	Rise: The Vieneo Province
Internal Account Name	Rise: The Vieneo Province
Legal Name	Unistellar Industries, LLC
Destination and image	Existing Rise homepage and station-model creative. The English ad's destination remains https://www.risetvp.com/ without the tracking parameters that were added to the archived translations.
Readiness at this update	The English ad is approved; the account is in review, and the web settings still report that billing setup is required. An active campaign does not establish that ads are serving.

Language caveat: “English-only” describes the manually saved ad set and the selected markets. The account's Creative enhancement setting was also observed to be On; its web description says it can translate ad content. Archiving the nine translated ads does not by itself establish English-only rendered output. Review that setting before claiming a strict display-language restriction. This documentation update did not change any advertising settings.

Evidence: live campaign, ad, and account reads, plus saved General settings in the web interface, on 23 October 2026. Execution and verification receipts are retained privately; the public record omits account identifiers and payment details.

What the earlier advertising taught us

The 2019–2020 notes above repeatedly record clicks without recorded conversions. The Facebook account paid closer attention to visitor interaction and new players. The useful lesson is to follow the journey into the game. Those notes do not document sufficiently consistent attribution, conversion definitions, or independent reporting periods to establish a reliable cross-platform acquisition-cost comparison. The earlier suspicion about automated Google traffic remains an historical interpretation, not a verified finding adopted by this experiment. (Historical Advertising record)

The old Facebook weekday findings and lookalike-audience results are useful context, but do not establish which days or targeting choices will work in ChatGPT in 2026. Keep a simple starting configuration and evaluate new evidence before changing scheduling, audience scope, or spending.

The promise and the initial settings

The creative follows the Gameplay positioning review: operating vehicles, understanding a physical world, and contributing to its development. Bill's accounts connect flying with construction, supply problems, and pride in completed colonies. The streamer feedback relayed by Jason identifies atmospheric simulation, persistence, and colony gameplay as particular strengths. These are reasons to test a focused promise; one player's enthusiasm does not measure the size of the market. (Gameplay source and decision record)

The following original launch baseline was saved and read back earlier on 23 October 2026. Preserve it as history; see Current setup for the later country, language, and branding decisions:

Item	Initial configuration
Campaign / ad group / ad	Risetvp campaign / Recommended ad group / Vieneo Colony
Objective and bidding	Clicks; automatic maximize clicks bidding; billed for clicks.
Audience scope	United States; no platform restriction configured. Mentioning Windows in a relevance hint does not enforce Windows-only delivery.
Budget	US$25 average daily budget, reduced from US$100 at Jason's request. Account timezone: America/Chicago. No campaign end date or total campaign budget is set.
Ad headline	Rise: Fly. Supply. Build.
Ad text	Master atmospheric flight. Supply colonies. Shape a persistent world.
Destination	Rise homepage
Initial image	Existing square image of a station model on a white background, retained with its full-image crop.

The headline and text replace “Join Vieneo” and “Explore a star system built for colonists.” The revised promise gives the prospective player a more concrete activity and purpose.

The five saved relevance hints are reproduced exactly:

1. Orbital maneuvers, atmospheric reentry and flight instruments connect Rise with an interest in learning detailed flight procedures.
2. Cargo and passenger transport link surface settlements with orbital facilities, giving logistics operations a practical purpose.
3. Player-built colonies produce resources and refine fuel, connecting transport work with trade and economic development.
4. A persistent multiplayer world supports long-term fleets, settlements and cooperative projects.
5. Rise requires Windows 10 or later; free in-game flight simulators let newcomers practise without owning an aircraft.

These hints describe relevant connections between a conversation and Rise. They are guidance for matching, not a list of exact-match search keywords or a guarantee of an audience. The initial audience hypothesis includes flight, transport, logistics, and settlement-building interests; it does not assume that every space-game fan wants Rise's pace and commitment. (Ads in ChatGPT: the basics)

Creative improvement to test: an authentic gameplay image showing an aircraft or cockpit approaching a settlement through weather would connect the visual more directly with the promise. This is a recommendation; the station image has not been replaced. Similarly, review the homepage-to-download-to-first-flight path so a newcomer can find free simulator practice and understand demo eligibility. The ad must not imply that colony ownership or every kind of work is available immediately to a demo player.

Budget, launch status, and the test period

US$25 was the documented minimum USD daily budget when checked. It is an average daily budget: a full unchanged Sunday–Saturday budget week allows US$175, and a day can allow up to twice the daily budget. A midweek change prorates the remaining weekly limit. On a change day, the daily maximum uses the highest budget active that day: the earlier US$100 setting means 23 October can retain a US$200 daily maximum, despite the reduction. Subsequent full days at US$25 have a US$50 daily maximum. These are limits, not spending forecasts. (Daily budgets)

Status at the initial 23 October check (historical): campaign, ad group, and ad were active, and the ad copy was approved. Delivery was blocked by the missing account logo and an absent or unauthorized payment method. Downstream delivery checks were skipped because of those account issues; this was not a complete launch clearance. No meaningful results were available for evaluating the experiment.

Start the evaluation clock when delivery actually begins. Review the first full budget week of delivery, then allow the associated player cohorts enough time to evaluate seven-day returns. This is a review recommendation, not an automatic pause, scheduled review, or total experiment cap. Whether ChatGPT can find enough suitable people, and whether they become players, remains to be measured.

Language targeting: what the translation trial established

On 23 October, nine translations—Dutch, French, German, Greek, Italian, Spanish, Brazilian Portuguese, Japanese, and Korean—were created alongside English, and the campaign was expanded to 26 countries. This was an approved setup trial; it did not establish demand or successful delivery in those markets.

There was no ad-to-country language map. All ten ads shared one ad group and the campaign's country list. Naming an ad “French” or “Japanese” did not constrain its geography. The connected controls exposed country targeting at campaign level, with no per-ad language or country override. Automatic relevance selection was not verified to guarantee a language match. OpenAI likewise describes context hints as relevance guidance, not targeting rules. (Campaign targeting; Context hints)

Jason then chose the simpler starting point: archive all nine translations, keep the English ad active, restore the eight countries listed above, and retain US$25/day total. Those changes were applied and read back. The translations remain historical archived records, not active alternatives.

Separate geographically restricted campaigns could constrain where each creative is eligible. Country is still only a proxy for language, especially in multilingual countries. Separate ad groups or instructions in relevance hints do not provide the missing geographic restriction in the controls inspected. Revisit language-specific campaigns only when their cost, routing controls, and landing-page experience justify them.

Brand names and a portfolio of products

OpenAI distinguishes Account Name (internal organization), Brand Name (the public advertiser label), and Legal Name (billing and verification). They can be edited separately in Settings → General. A Brand Name change triggers brand review and stops serving during review; changing the legal name can restart business review. (Account, brand, and legal names)

The Rise ad initially displayed Unistellar Industries, LLC in its advertiser header. Renaming the campaign did not change that label. Jason requested Rise: The Vieneo Province as the public Brand Name while retaining Unistellar Industries, LLC as the legal name. Both values were verified in the saved web settings. The internal Account Name also now reads Rise: The Vieneo Province. The plug-in did not expose the separate public Brand Name field, so the web interface was used.

Multiple products do not inherently require multiple ad accounts. Products can share the company's advertiser identity while using their own ads and destinations. However, the inspected controls provide one public brand identity per account, with no campaign/ad-level override. Separate accounts were therefore discussed as a workaround for independently branded headers for Rise, SunTanApp, Overmonitor, AssetBug, NotifyPrice, and other products—not as a universal requirement or an approved portfolio rollout.

Account limits, minimum budgets, and the expansion decision

Eligible businesses can create additional advertiser accounts under the same OpenAI login. Self-service creation stops once that login belongs to ten or more ad accounts; both created accounts and invited memberships count. This is a creation restriction, not a prohibition on membership in more than ten existing accounts. Eligibility also depends on billing and organization permissions. (Additional accounts and creation limits)

The documented USD minimum is US$25 for each daily-budget campaign. It is not a per-ad minimum, an account subscription price, or a guaranteed charge. The minimum should not be generalized to campaign-total budgets. (Minimum campaign spend)

Example with a US$25 daily budget per campaign	Combined average daily budget setting
Current Rise experiment: one campaign	US$25
Ten simultaneous daily-budget campaigns, one per language	US$250
Fifteen simultaneous daily-budget campaigns, one per product	US$375

These are arithmetic budget examples, not spending forecasts or approved expansions. The number of accounts alone does not determine the budget. The existing campaign's one US$25 budget was shared across all ads during the translation trial; creating separate daily-budget campaigns would introduce separate minimums.

Jason identified the administrative burden and multiplied minimum budgets as a potential deal-breaker for a portfolio of roughly fifteen products. The adopted configuration remains the single English-ad Rise experiment at US$25/day. The working recommendation is to measure player acquisition and retention before expanding, and to reassess whether the platform's branding, language controls, and economics fit the broader portfolio. No fifteen-account structure or US$375/day spend was authorized.

How we will judge it

The following measurement work is proposed, rather than reported as already implemented:

1. Establish attribution. Verify how an ad visit connects to a download, a new account, and subsequent play. No conversion event is selected on this campaign, and no campaign or ad-group URL-tracking template was configured at review. That does not establish that website or game tracking is absent. Audit the existing linkage before reporting advertising-attributed players.
2. Follow the whole journey. Record impressions, clicks, spend, and cost per click alongside engaged visits, downloads, new players, a defined first meaningful flight or task, and returns after seven days. Keep downloads, accounts, active players, and retained players distinct. Define a return consistently and report acquisition dates and the number eligible for follow-up.
3. Diagnose where people stop. If clicks do not lead to players, check attribution, Windows compatibility, landing-page expectations, installation, and the route into training. Observe newcomers and distinguish independent progress from mentor-assisted progress, as recommended in Gameplay. A weak result alone does not identify bots or prove there is no audience.
4. Earn any budget increase. Assess cost per attributable new player and per returning player, together with first-session completion and player feedback. Establish an affordable acquisition cost from Rise's economics before scaling. At small numbers, report counts and uncertainty. Click-through rate or cheap clicks alone do not establish success.

Ads Manager reporting supplies delivery and interaction measures; conversion reporting depends on its configuration and attribution rules. Platform conversion totals must not automatically be treated as unique new players. This initial setup is an observational acquisition test, not a controlled comparison proving that the new copy outperforms the previous copy. (Measure results)

Decision history and attribution

Jason requested the campaign tuning after the comparison and Gameplay discussions, approved the revised copy and relevance hints on 23 October 2026, then explicitly chose US$25 per day and requested this experiment record. Codex applied and verified those changes through Ads Manager and prepared this section. The measurement plan, creative improvement, and scaling criteria are recommendations for evaluation; they are not claims of completed implementation or adopted game features.

The later language, branding, account-limit, and budget lessons were added at Jason's request on 23 October 2026. They distinguish his approved choices from platform documentation, observations of the available controls, and recommendations. The translation trial and its rollback are recorded as configuration changes, not performance results.

The baseline comes from live Ads Manager reads on 23 October 2026, with execution receipts retained privately. Design rationale comes from Gameplay revision 4991 and its attributed sources, including Bill's historical testimony and the streamer comment relayed by Jason. Platform mechanics come from the official Help Center pages linked above, reviewed for this setup. Future results should add dated entries with delivery dates, spend, attribution method, player counts, and any settings changed, preserving this starting record.